Order the axes by the F statistics from analyses of variance based on a grouping variable (as illustrated in body dataset)
# data(body, package="gclus")
# body1 <- body
body1$Gn <- factor(body1$Gn)
ggparcoord(body1, columns=1:24, scale="uniminmax",
alphaLines=0.4, groupColumn="Gn",
order="allClass") + xlab("") + ylab("") +
theme(legend.position = "none",
axis.ticks.y = element_blank(),
axis.text.y = element_blank())

- we can display two graphics, one for women and one for men, to avoid overplotting for the variables where the genders overlap.
a <- ggparcoord(body1[order(body1$Gn),], columns=c(1:24),
groupColumn="Gn", order="allClass",
scale="uniminmax") + xlab("") + ylab("") +
theme(legend.position = "none",
axis.ticks.y = element_blank(),
axis.text.y = element_blank()) +
scale_colour_manual(values = c("grey","#00BFC4"))
b <- ggparcoord(body1[order(body1$Gn, decreasing=TRUE),],
columns=c(1:24), groupColumn="Gn", order="allClass",
scale="uniminmax") + xlab("") + ylab("") +
theme(legend.position = "none",
axis.ticks.y = element_blank(),
axis.text.y = element_blank()) +
scale_colour_manual(values = c("#F8766D","grey"))
grid.arrange(a,b)

Order the axes by their median values (body dataset)
##
## Attaching package: 'dplyr'
## The following object is masked from 'package:car':
##
## recode
## The following object is masked from 'package:gridExtra':
##
## combine
## The following objects are masked from 'package:stats':
##
## filter, lag
## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, union
m2 <- apply(body[, 1:24], 2, median, na.rm=TRUE)
m2a <- order(m2)
ggparcoord(data = select(body, -Gender), alphaLines=0.3,
scale="globalminmax", order=m2a) + coord_flip()

Order the axes by their maximum values after standardization by mean and standard deviation (body dataset)
B1 <- ggparcoord(data = body1, columns=c(1:24), scale="std")
B2 <- acast(B1$data[ ,c(2,4,5)], .ID ~ variable)
m4 <- apply(B2, 2, max, na.rm=TRUE)
m4r <- order(m4)
ggparcoord(data = body1, alphaLines=0.3,
columns=c(1:24), scale="std", order=m4r)
